Verdonschot E H, Wenzel A, Bronkhorst E M
Department of Cariology and Endodontology, TRIKON, The Netherlands.
Community Dent Oral Epidemiol. 1993 Oct;21(5):269-72. doi: 10.1111/j.1600-0528.1993.tb00772.x.
Accuracy of diagnostic tests for caries detection is frequently evaluated by sensitivity and specificity values obtained from data acquired on an ordinal rating scale which classifies lesions according to the depth in enamel and dentin. The interpretation of these parameters may, however, be difficult. Receiver Operating Characteristic (ROC) analysis, on the other hand, yields an expression for overall diagnostic accuracy in a single parameter, facilitating a comparison of performance of several diagnostic tests. The basis for applying ROC analysis has been that data were collected on a confidence (likelihood) scale. The aim of this preliminary study was to investigate whether the application of ROC analysis would yield as valid parameters for diagnostic accuracy in assessment of approximal caries on a depth rating scale as on a likelihood rating scale. The results demonstrated that the accuracy of observer performance was not influenced by the type of rating scale employed, justifying the application of ROC analysis on discrete caries depth ratings.
龋齿检测诊断测试的准确性通常通过灵敏度和特异度值来评估,这些值来自于按牙釉质和牙本质深度对病变进行分类的有序评分量表所获取的数据。然而,对这些参数的解释可能会很困难。另一方面,受试者工作特征(ROC)分析会得出一个单一参数的总体诊断准确性表达式,便于比较几种诊断测试的性能。应用ROC分析的依据是数据是在置信(似然)量表上收集的。这项初步研究的目的是调查ROC分析的应用在深度评分量表上评估邻面龋时,是否能像在似然评分量表上一样得出有效的诊断准确性参数。结果表明,观察者表现的准确性不受所采用评分量表类型的影响,这证明了在离散的龋病深度评分上应用ROC分析是合理的。